26 M a y 2017 V o l u m e b y G l a c i e r R e s e a r c h. Inflation-linked bonds An introduction

Transcription

1 FUNDS ON FRIDAY b y G l a c i e r R e s e a r c h 26 M a y 2017 V o l u m e 9 10 Bootstrapping the term structure of priced-in future inflation from nominal and inflation-linked bonds Written by: Andreas Tindlund, Senior Manager Fixed Income at Terebinth Capital Inflation-linked bonds An introduction Nominal South African government bonds pay a fixed coupon semi-annually and the principal back at maturity. The cash flow for the security is therefore known. Inflation-linked bonds (ILBs) differ from nominal bonds in this regard. The principal amount of the ILB grows every month with the rate of monthly headline inflation as released by Stats SA. Coupons are paid semi-annually, whereas on coupon payment dates the fixed real coupon is paid out on the current principal value of the bond. Thus, unlike a nominal bond, the cash flow stream on ILBs is unknown ex-ante as both coupons and the final principal amount at maturity will grow in line with inflation over the life of the bond. Because the coupon rate on ILBs is much lower than that of nominal bonds, the cash flow is also significantly more back-loaded, which means that the duration of an ILB compared to a matched-maturity nominal bond is higher, with the ratio growing the longer dated the bonds are. To illustrate this, Figure 1 below shows a comparison of the fixed cash flows of R100 notional in the R2032 (8.25% March 2032 maturity) and a simulation of the cash flows of R100 notional in the I2033 (1.875% Feb 2033 maturity) assuming 5.8% CPI straight-line inflation until maturity. As can be seen, the coupons from the ILB are significantly smaller, but growing, over the life of the bond, while the principal payment at maturity is 2.7 times that of the nominal bond. Even though the years to maturity of the nominal and ILB are similar at 15.0 and 15.9 respectively, the duration of the nominal bond is only 8.11 compared to the ILB at This is because the most significant Glacier Financial Solutions (Pty) Ltd, l A member of the Sanlam Group Private Bag X5 l Tyger Valley 7536 l client.services@glacier.co.za Tel l Fax l Web Reg No 1999/025360/07 l Licensed Financial Services Provider This document is intended for use by intermediaries. The information contained in this document has been recorded and arrived at by Glacier Financial Solutions (Pty) Ltd in good faith and from sources believed to be reliable, but no representation or warranty, expressed or implied, is made as to its accuracy, completeness or correctness. The information is provided for information purposes only and to assist the financial intermediary to submit an investment proposal to the client and should not be construed as the rendering of investment advice to clients. Page 1

2 cash flow of the ILB is the principal payment at maturity, in this simulation at R270. Figure 1 Fixed cash flows of R2032 and a simulated cash flow of I2033 assuming 5.8% straight-line future inflation Consumer price inflation (CPI) in South Africa has varied greatly over recent history, and has averaged 5.8% since the beginning of The Reserve Bank s mandate is to target inflation within a 3% to 6% band. Figure 2 Year-over-year inflation Bootstrapping fair value zero-coupon yields for SA government fixed cash flows The current outstanding fixed-rate nominal bonds where rand cash flows are fixed allow us to bootstrap a term structure for discounting factors or zero-coupon yields for South African Government Bond cash flows. The zerocoupon yields are shown in the figure below in blue. The corresponding derived continuous forward yields are shown in orange. In a risk-neutral world, this line would be equal to the expected path of bond financing overnight repo rates over the next 30 years, but in real life it incorporates in addition term-, inflation-, and credit- and liquidity premiums. By extending an investment in a nominal bond from for example an existing 10-year nominal bond to a 15-year nominal bond, one would lock in the forward yield of the orange line of around 11.6% from year 10 to 15. Page 2

3 Figure 3 Bootstrapped zero-coupon yields and continuous forward yields exhibited by nominal bonds Using the bootstrapped curve for nominal South African Government Bonds we can further boot-strap the future path of inflation priced in to ILBs. This is done by modelling the CPI path and corresponding ILB cash flows discounted at the zero-coupon yields exhibited by stock of nominal bonds. In a risk-neutral world this would be the expected path of inflation, but in real life it incorporates inflation- and liquidity premiums. The term and credit premiums are already incorporated into the zero-coupon curve given by the nominal bonds. ILBs are much less liquid than nominal bonds and so should call for a higher liquidity premium, but unlike nominal bonds they should have no inflation risk premium. Figure 4 Market implied forward path of inflation, and the zero-coupon inflation breakeven term structure The result of this exercise can be seen in the figure above. The orange line shows the forward path of YoY inflation priced in by ILBs. If CPI follows this line in years to come, return on the ILBs will produce the same investment Page 3

4 return as similar nominal bonds, (assuming all coupons paid can be reinvested at the current forward yields in the bootstrapped nominal curve). The shortest ILB is the R212 which matures in five years, and implies on average a steady 5.8% inflation over this period, in line with the historical average. From this point forward however, the ILB market prices in significantly higher inflation reaching close to 9% in year eight. From year 15 further higher inflation is priced in at 9.65% for the rest of the horizon. The blue line in the chart is the zero-coupon inflation priced in by the blue inflation path from today. This shows the true term structure of breakeven inflation. Break-even inflation is defined as the inflation required over the period for an ILB to produce the same return as a nominal bond. For example, as shown by the blue line in the figure above, it requires an average compounded inflation rate of just over 7% from today over the next 10 years for a 10- year ILB to break even against a 10-year nominal bond. In the market, breakeven inflation is commonly calculated by a more simplistic measure, namely the difference between the yield on nominal bonds and the real yield of ILBs of similar maturity. This is very easy to calculate, however, it is not accurate. As discussed, the cash flows and duration for similar nominal and inflation-linked bonds are vastly different, and are more pronounced for high-inflation countries like South Africa, compared to lowinflation developed markets which have low coupon bonds. The figure below shows conventional break-even (difference between ILB real yield and interpolated nominal yields) compared to the true break-even (zero-coupon inflation) yields bootstrapped above. As can be seen, with the current steep term structure for nominal bonds, the true break even yield is significantly higher than the one computed by the conventional method, with longer-dated bonds most affected. For the longest dated bonds for example, it will require an average annual inflation rate of 8.63% from today for this investment to break-even compared to 7.40% suggested by the conventional method. This difference is massive when considering it compounds for 30 years. Figure 5 Comparison of conventional and true zero-coupon inflation breakeven rates. Page 4

5 If we assume expected future inflation equal to the historical average of 5.8% across the horizon and no inflation risk premium, the fair value price for the longest dated ILB, I2050, would be R compared to the current price of , a full 38% lower, and a real yield of 4.61% compared to the current 2.34%. Various countries issue 30yr ILBs. Looking at the real yield offered on these securities vs the countries credit rating shows a not surprising relationship where lower ratings are correlated with higher real yields. On this metric, however, South Africa stands out as offering very low real yields compared to its low credit rating as shown in Figure 6. Based on its current rating, a long-dated real yield closer to 4% would be in line with peers, from 2.3% currently. Figure 6 30yr ILB real-yields offered by various countries shown against composite foreign currency credit rating One of the reasons that ILBs, especially longer dated ones, appear overly expensive may be explained by supply/demand imbalances, where investment mandates are benchmarked to or require a certain allocation to ILBs, while issuance is not high enough to balance the market. On the supply-side National Treasury has been loath to issue a higher share of ILBs in its funding mix, despite that, to the author s clear view, this would significantly lower the cost of debt service of the country at current levels. One possible reason is the coupons and principal payments are unknown and therefore tricky to model and a fear that they can cause higher risk to national accounts should inflation accelerate. However, this fear is unwarranted, as the Treasury s revenues in terms of tax receipts and government assets move in line with inflation which will offset this effect. In fact, the risk to the repayment profile on the national debt lies with nominal bonds, as future potential disinflation will make the repayments much higher in real terms against lower nominal tax revenues. Page 5

6 Figure 7 Inflation-linked bonds have outperformed nominal bonds over the last decade. This is largely explained by the yield differential between nominal and inflation linked bonds (breakevens) have soared over the period. The spread between two old long-dated bonds, nominal R209 and ILB R202, went from 457bp at the start of the series to 740bp today In addition, and potentially partly due to the illiquid nature of ILBs, realised volatility is low as daily yield moves are small, with mark-to-market levels for most bonds not changing from one day to the next. This has coupled with a historical yield compression of SAGBs in both real and nominal terms, but with ILBs having outperformed nominals with lower volatility. This history of outsized returns with lower volatility appear favourably in efficient frontier models used in asset allocation processes inside investment houses and allocators which may explain higher demand for the product. Long-dated inflation-linked South African government bond valuations appear overly stretched and are likely set to deliver disappointing long-term returns for investors vis-a-vis nominal bonds going forward. Past outsized relative returns can only be repeated should inflation significantly breach on a sustained basis well outside the central bank s current mandate. Page 6

7 Appendix Bond Coupon Maturity Yield Cheap/(Rich) bp vs Bootstrapped curve Inflation-linked bonds Nominal bonds R / R / R / R / R / R / R / R / R / R / R / R / R / R / R / R / Bond Coupon Maturity Yield Cheap/(Rich) bp vs Bootstrapped CPI future path Assuming 5.8% flat CPI future path Cheap/(Rich) bp Current Price Fair price Diff R / % R / % I / % R / % I / % I / % R / % I / % I / % I / % Figure 8 Cheap/ (Rich) - Yield bp cheap (positive) or expensive (negative) vs bootstrapped nominal and inflation curves; Price date: 12 April 2017 Page 7

8 Glacier Research would like to thank Andreas Tindlund for his contribution to this week s Funds on Friday. Andreas Tindlund, CFA MSc Industrial Economics Andreas obtained his MSc degree from the Norwegian University of Science and Technology in 2006, completing parts of it at National University of Singapore and UCT. After graduation he moved to New York working for NBIM, the Norwegian sovereign wealth fund, as portfolio manager for the fund s $15bn North American fixed income credit portfolio. In 2008 he played an integral role in starting up NBIM s fixed income trading desk, taking up the position of trader and later senior trader. In 2011 he achieved his CFA charter, and in 2012 he became portfolio manager for the fund s $35bn global emerging market fixed income portfolio across South Africa and 19 other local markets. In 2015 Andreas joined the fixed-income macro focussed asset manager Terebinth Capital as senior portfolio manager. Page 8